maximkazhenkov

maximkazhenkov's Comments

Go F*** Someone

This is self-help-books-level advice

Is backwards causation necessarily absurd?

Yes, but the direction of causality is very much preserved. The notion of present is not necessary in a directed acyclic graph.

Predictors exist: CDT going bonkers... forever

But considering that randomness as an antidote to perfect predictions is ubiquitously available in this universe, it's hard to see what practical implications these CDT failures in highly contrived thought experiments have.

Plausible A.I. Takeoff Scenario Short Story
Any good idea can be enough for a successful start-up. AGI is extremely narrow compared to the entire space of good ideas.

But we're not comparing the probability of "a successful start-up will be created" vs. the probability of "an AGI will be created" in the next x years, we're comparing the probability of "an AGI will be created by a large organization" vs. the probability of "an AGI will be created by a single person on his laptop" given that an AGI will be created.

Without the benefit of hindsight, is PageRank and reusable rockets any more obvious than the hypothesized AGI key insight? If someone who had no previous experience working in aeronautical engineering - a highly technical field - can out-innovate established organizations like Lockheed Martin, why wouldn't the same hold true for AGI? If anything, the theoretical foundations of AGI is less well-established and the entry barrier lower by comparison.

Plausible A.I. Takeoff Scenario Short Story
I actually agree that the "last key insight" is somewhat plausible, but I think even if we assume that, it remains quite unlikely that an independent person has this insight rather than the people who are being paid a ton of money to work on this stuff all day.

If that were true, start-ups wouldn't be a thing, we'd all be using Yahoo Search and Lockheed Martin would be developing the first commercially successful reusable rocket. Hell, it might even make sense to switch to planned economy outright then.

Especially because even in the insight-model, there could still be some amount of details that need to be figured out after the insight, which might only take a couple of weeks for OpenAI but probably longer for a single person.

But why does it matter? Would screaming at the top of your lungs about your new discovery (or the modern equivalent, publishing a research paper on the internet) be the first thing someone who has just gained the key insight does? It certainly would be unwise to.

Phage therapy in a post-antibiotics world

Seems unlikely as phages can evolve just as fast.

What would happen if all the water on Earth were accumulated into spheres & drop on the surface?

These follow-up questions pertain to a dynamic context, and I'm afraid I'm not equipped to answer them. Moreover, I would also claim that not even Randall Munroe himself would be able to answer these questions, or anyone who hasn't got a supercomputer and a team of physicists at disposal.

I bought the What If book myself and loved every chapter of it. But if you look closely, you will notice that basically every analysis in that book was made from a static context or a dynamic one that has ridiculously simple solutions (i.e. linear or exponential). Even exotic topics like neutron star matter and supernova neutrinos can be analysed with ease under a static context; just a matter of typing large numbers into a calculator. But as soon as dynamics is involved, even mundane things like Earthly weather or air flow over ailerons is going to require a supercomputer.

It also doesn't help to analogize the problem with more familiar scenarios, either. Quantity has a quality of its own, as Stalin famously said. Thing like the cube-square law make big things behave very differently than small things even if they're made out of the same material or undergoing the same basic process. Nuclear explosions and supernovas are not hard to understand because of the extreme energies involved per se. Nuclear interactions relevant to these processes are many orders of magnitude lower than the energies achieved in particle accelerator experiments. What macroscopic effect a gargantuan amount of these simple interactions can produce, however, is a different matter.

That's why you need lots of brute force computational power as well as a team of physicists doing clever simplifications to get a general understanding of the problem at hand, not even a precise prediction of a specific problem instance like weather forecast. And I'm afraid they won't let you borrow their precious compute for a fun thought experiment.

Worse yet, in the case of real phenomenons like nuclear explosions and supernovas we at least get to observe their aftermaths (bomb yield/supernova remnant) to set a few boundary conditions on our analysis. For completely hypothetical scenarios, we can't even check our predictions against reality. Can we, for instance, safely ignore temporary phase changes into exotic ice forms? How about nuclear interactions triggered by locally extreme heat and pressure?

Tapping Out In Two

If you'd like to break the habit of getting into internet arguments in the first place, this might be the right thing (for YouTube):

chrome.google.com/webstore/detail/hide-youtube-comments

The unexpected difficulty of comparing AlphaStar to humans

That makes sense. Perhaps the opposite is true - that if all Nash equilibrium strategies are mixed, the game must have been imperfect information? In any simultaneous game the opponent's strategy would be the hidden information.

Load More